AI Model Discovers New Physics in Many-Body Systems
Researchers at Emory University are using AI to discover new principles in the physics of many-body systems. The work provides a universal framework that could accelerate research across hard sciences and create commercialization opportunities for deeptech startups.
- The specific physical system studied was dusty plasma, an ionized gas containing suspended microparticles, which serves as a model for more complex many-body systems. - The AI, a physics-based neural network, described the non-reciprocal forces between particles with over 99% accuracy, correcting prior theoretical assumptions that did not fully account for particle size. - Senior co-authors on the PNAS-published paper are Emory professors Justin Burton (experimental physics) and Ilya Nemenman (theoretical physics), with Wentao Yu, now at Caltech, as the first author. - Unlike a "black box," the researchers emphasize their AI's framework is universal and understandable, making it applicable to other systems like biological cells, industrial colloids such as paint, and even collective behaviors like flocking birds. - The commercialization of such university research often follows a path where AI-driven simulations and experiment optimization lead to the discovery of new materials or therapies, forming the basis for deeptech startups. - In Turkey's deeptech ecosystem, there are 58 AI-focused companies that have collectively raised $17.3M; funding for this sector saw a 168% increase in 2024 compared to the previous year. - Key Turkish venture funds focusing on deeptech include Diffusion Capital Partners (DCP) and TechOne VC, which invest in early-stage technology startups with global potential. - Turkish deeptech startups are also gaining international exposure through programs like the NATO DIANA accelerator, which selected Istanbul-based Qubitrium for its first cohort to advance its work in quantum cryptography and sensing.
